OpenAI opened ChatGPT Health to users across the United States on Thursday (23), extending the service beyond medical professionals to the broader public. At the briefing, Ashley Alexander, the company’s vice president of health products, said OpenAI’s models now have reasoning ability that “exceeds the level of clinical physicians.”
That claim landed in the same week OpenAI was hit with a lawsuit in San Francisco Superior Court. Scott Winters, a 55-year-old former pastor from Florida, alleges that ChatGPT gave “extremely dangerous medical advice,” causing him to delay seeking care until a blood clot developed into a massive pulmonary embolism.
From clinician-only access to a nationwide consumer rollout
The change is not a brand-new product launch but a broader release. On April 23 this year, OpenAI introduced ChatGPT for Clinicians, which was limited to verified medical professionals, and released the HealthBench Professional evaluation at the same time. That version targeted physicians rather than the general public.
Three months later, OpenAI removed that gate. The service is now available to U.S. users aged 18 and older who are logged into ChatGPT, covering Free, Go, Plus and Pro plans. The initial rollout is on the web and iOS.
OpenAI said the expansion is backed by GPT-5.6 Sol, a model released a few weeks earlier. The company described it as its strongest health model to date, and said ChatGPT Health will run on it going forward.
The company also folded health queries into the main chat experience. Previously, users had to switch to a separate ChatGPT Health tab to ask about health records and related information. Now those questions can be asked directly in the primary chat window.
Health data access and encryption claims
Before use, ChatGPT asks for permission to use health data in personalized responses. OpenAI said all conversations are encrypted both in transit and at rest, and that data linked into Health receives additional encryption protections.
Users can review and adjust the types of linked information, including post-visit summaries, care team notes, lab reports and medication records. Data sources include medical providers as well as third-party services such as Apple Health, Weight Watchers, MyFitnessPal and Function.
The company’s privacy pitch focused on encryption. The dispute raised this week, though, centers on something else as well: what happens after the system produces advice, and whether users trust it enough to act on it.
Lawsuit filed one day before the public launch
According to the complaint, Winters sued OpenAI in San Francisco Superior Court on July 22. He claims ChatGPT-4o repeatedly minimized his symptoms, told him to stay home and rest in a recliner, and discouraged him from going to a doctor. He says that delay allowed a blood clot to worsen into a massive pulmonary embolism.
The complaint also alleges that ChatGPT used Winters’ own religious background to persuade him, framing its advice in faith-based language that made him more comfortable staying home instead of going to the emergency room.
The suit raises eight causes of action, including negligence, alleged violations of California consumer protection law and privacy law, and a direct claim that OpenAI engaged in the unlicensed practice of medicine while prioritizing user engagement over safety. In plain terms, the plaintiff argues that ChatGPT’s responses went beyond informational reference and effectively substituted for medical diagnosis and treatment decisions.
It remains unclear which model Winters actually used.